Justin Johnsen, KPMG’s lead technical architect, says his role as a forward-deployed engineer has changed over the past year from software writing to client-facing AI work. He has been at KPMG for six years and says he has spent about a year working across different clients in this role.
Johnsen says a forward-deployed engineer works directly with a client and stays with the problem from understanding the business need through delivering a working solution. In practice, that means being on-site with client teams, gathering requirements, and helping turn those needs into software and AI-driven deliverables.
A big part of his work now is AI. Johnsen says AI is 'incredibly capable' but only as good as its context. He says much of his time goes into gathering the information AI needs to generate useful assets and artifacts, then spreading those outputs across client deliverables and software applications. He also coaches client teams on how to use AI effectively.
He gives one example where he was hired to integrate several applications into a complex software system. After months of gathering requirements and refining the context, his team used AI to deliver the application in about a month. The client, he says, wanted to understand how the work moved so quickly and how transparent the process had been. That same client then asked him to help its teams adopt the AI and prompting practices behind the result.
Johnsen says AI has made it easier to multiply his output and 'act as my own enterprise.' The bottleneck used to be software development and writing code, he says; now it is gathering context and refining intent before development begins.
He also describes a recent application involving risk scoring, where AI was used to assess how risky a customer or partner might be for a client to do business with. His team brought together the signals used to measure that risk, then used AI to reason across them and generate a score that independent reviewers could use in their assessments.
The broader shift, Johnsen says, is that more software-development work is moving toward the data and integration layers. Many enterprise problems now revolve around helping AI make sense of data, bringing information together for monitoring and visibility, and connecting legacy and cloud applications that do not naturally communicate with one another.
For founders, the playbook is familiar: the value is no longer in raw output alone, but in the systems, context, and adoption work around it. AI can speed delivery, but only if someone owns the messy middle between the business need and the finished product.
That is where the leverage sits. The firms that win will not just ship code faster; they will turn client knowledge, integrations, and training into a repeatable engine. In a market where execution matters, the bottleneck has moved upstream.
